Re: UTC, GMT, leap seconds
> POSIX mandates that system time calculations do not account for leap
Correct, POSIX basically mandates being "in denial" about leap
seconds. The second can actually occur (tm_sec can equal 60... or 61,
but that's a typo in the spec, IERS would not do that unless they'd
already done 365 single-day leap seconds that year, and if we're
losing 6 min/year, leap seconds are *not* our biggest problem :-)
Note that debian includes a set of "right" timezone values, so you can
see how/if it matters:
% env TZ=right/US/Eastern stat /etc/motd
File: "/etc/motd"
Size: 403 Blocks: 2 Regular File
Access: (0644/-rw-r--r--) Uid: ( 0/ root) Gid: ( 0/ root)
Device: 302 Inode: 36908 Links: 1
Access: Wed Jul 28 22:16:49 1999
Modify: Fri Jul 16 13:48:21 1999
Change: Sun Jul 18 22:59:38 1999
% env TZ=US/Eastern stat /etc/motd
File: "/etc/motd"
Size: 403 Blocks: 2 Regular File
Access: (0644/-rw-r--r--) Uid: ( 0/ root) Gid: ( 0/ root)
Device: 302 Inode: 36908 Links: 1
Access: Wed Jul 28 22:17:11 1999 <<< \
Modify: Fri Jul 16 13:48:43 1999 <<< -* note the differences, 22 leap seconds
Change: Sun Jul 18 23:00:00 1999 <<< /
Note that the use of right/* has been known to break packages like
Kerberos in the past, due to either libc bugs or inconsistently
converting one way with libc and back with non-libc code that followed
the POSIX algorithm (I don't know if Kerberos V5 handles this
correctly yet.)
_Mark_ <eichin@thok.org>
The Herd of Kittens
Debian Package Maintainer
Reply to: